|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|
The results indicated that the IS wclay-bearing sand reservoirs have been found to be very relationship was generally complicated due to the additional conductivity of the clay a nonlinear relationship, which changed with the water minerals, which tend to increase with the increase of porewater saturation on a log-log plot. It was observed that the clay conductivity in a nonlinear relationship. Therefore, volume fractions, as well as the clay minerals in which the the aim of this study is to investigate the effects of porewater conductivity varied with the pore-water conductivity, had conductivity and clay minerals on the saturation displayed signicant effects on the saturation exponent.
Saxena, Nishank (Shell International Exploration & Production) | Hows, Amie (Shell International Exploration & Production) | Hofmann, Ronny (Shell International Exploration & Production) | Alpak, Omer (Shell International Exploration & Production) | Freeman, Justin (Shell International Exploration & Production) | Appel, Matthias (Shell International Exploration & Production) | Dietderich, Jesse (Shell International Exploration & Production)
Digital Rock Physics or Digital Rock is a rapidly advancing image-based technology for predicting subsurface properties of complex rocks (e.g., porosity, permeability, formation factor) to achieve more, cheaper, and faster results as compared to conventional laboratory measurements. We show that a combination of limited image resolution and a finite field of view, leads to the systematic underestimation of porosity and overestimation of permeability calculated using Digital Rock technology. We quantify these effects and derive the necessary corrections to estimate the true rock properties directly from the segmented micro-CT image. These novel solutions allow us to further mature DRP as a technology for existing and future fields.
Digital Rock Physics (DRP) aims to predict subsurface properties of complex rocks to achieve more, cheaper, and faster results as compared to conventional laboratory measurements (Fredrich et al., 1993; Keehm et al., 2001; Arns et al., 2005; Knackstedt et al., 2009). This is possible since DRP analyses can be carried out on a “as received” piece of rock (< 1 cm3) that can come from whole core, sidewall core, or a drill cutting, using stateof-the-art micro-CT technology (Figure 1) and a combination of novel digital approaches (e.g., direct numerical simulation, deep learning, machine vision). However, for this technology to mature, it is important to demonstrate that it is feasible to estimate two of the most fundamental of all rock properties, porosity and permeability. Shell has recently performed a series of detailed studies to benchmark commercially available Digital Rock tools (Saxena et al., 2017) and vendors and arrived at the conclusion that there is no viable external commercial solution, as of 2018, that provides sufficiently accurate results in comparison to laboratory measurements. Raw image-derived properties typically underestimate porosity (by up to 6 p.u.) and overestimate permeability (by a factor of 10 or more) for reservoir rocks in Shell’s portfolio. This is because the current pore-scale imaging is limited to image resolution of 1 - 2 microns and field of view of about 20003 - 40003 voxels.
Digital rocks obtained from high-resolution micro-computed tomography (micro-CT) imaging has quickly emerged as a powerful tool for studying pore-scale transport phenomena in petroleum engineering. In such frameworks, digital rock analysis usually carries the problematic aspect of segmenting greyscale images into different phases for quantifying many physical properties. Fine pore structures, such as small rock fissures, are usually lost during segmentation. In addition, user bias in this process can lead to significantly different results. An alternative approach based on deep learning is proposed. Convolutional Neural Networks (CNN) are utilized to rapidly predict several porous media properties from 2D greyscale micro-computed tomography images in a supervised learning frame. A dataset of greyscale micro-CT images of three different sandstones species is prepared for this study. The image dataset is segmented, and pore networks are extracted to compute porosity, coordination number, and average pore size for training and validating our model predictions. The greyscale images (input) and the computed properties (output) are uploaded to a deep neural network for training and validation in an end-to-end regression scheme. Overall, our model estimates porosity, coordination number, and average pore size with an average error of 0.05, 0.17, and 1.8μm, respectively. Training wall-time and prediction error analysis are also discussed. This is a first step to use artificial intelligence and machine learning methods for the robust prediction of porous media properties from unprocessed image-driven data.
Summary to handle regular and irregularly shaped grains, 4) handles multiple grain types at once, 5) has the ability to deposit We developed a numerical framework and procedures to grains using multiple scenarios (e.g. An efficient approach to create multiple realizations of nonspherical irregularly shaped grains using Generation of irregular grains coherent noise modification of the spherical grains surface is introduced. Various three-dimensional random loose and Figure 1 illustrates the numerical procedure for creating nonspherical dense grain packs show the validity, flexibility, and irregular grains. Correlated noise is used to consistency of the simulator compared to previous studies. The procedure consists of the following steps: 1) Introduction generating a spherical mesh using regular convex icosahedron approximation (Popko, 2012), 2) generating Studies on granular media are essential for understanding a three-dimensional coherent noise cube, 3) displacing the wide range of physical phenomena including fluid flow, spherical mesh by adding the values of the scaled threedimensional stress and strain, heat conduction, and electrical effects.
Since the 1990s, X-ray computed tomography (CT) has become increasingly popular for its ability to yield big insights from tiny rock samples. But the volume of data produced from CT scanners is anything but tiny, and that makes sharing and storing the files a challenge. In an effort to address this issue and foster collaboration in an area where there is currently very little, researchers at the University of Texas at Austin (UT) created a new web-based application called Digital Rocks Portal. In exchange for making their CT images of rocks available to outside researchers, the website will offer exploration com panies free and unlimited data storage. Maša Prodanović, the director of the Digital Rocks Portal and an assistant professor at UT's petroleum and geosystems engineering department, said this arrangement will improve the industry's understanding of rock microstructures and the accuracy of microscopic hydrocarbon flow models--valuable information that can be extrapolated to the reservoir scale.
Due to limits of currently technology, it is a great challenge to research the dispersion of acoustic velocity in middle frequency domain with only rock experiments. In this research, digital rock samples were constructed by piling up of both the spherical and the cubic grains according to volume fraction of pore space. Lattice Boltzmann method (LBM) has been further extended to investigate the acoustic transport properties in porous media saturated with fluid. The validity of LBM can be verified by comparing the simulation results with the theoretical calculation of Voigt-Reuss model and Hashin-Shtrikeman model. The dependence of the acoustic properties on frequency has been investigated by LBM to reveal the relationship between velocity and frequency under different conditions. Numerical results show that an obvious change of acoustic velocity with frequency could be observed in each of the samples. Therefore, a new model has been developed to formulate the relationship in a wide range of frequency from a few Hz to thousands Hz, which may lay the theoretical foundation for analysing reservoir through combining different geophysical methods.
Presentation Date: Monday, September 25, 2017
Start Time: 3:55 PM
Location: Exhibit Hall C/D
Presentation Type: POSTER
Saxena, Nishank (Shell International Exploration & Production) | Saenger, Erik H. (International Geothermal Centre, Bochum University of Applied Sciences) | Hofmann, Ronny (Shell International Exploration & Production) | Wiegmann, Andreas (Math2Market GmbH)
Digital Rock Physics (DRP) technology can improve calibration of empirical and theoretical rock physics models in the absence of direct core measurements. To this end, we must quantify the degree of variability associated with computation of elastic and sonic properties using various numerical engines. To add to this ongoing discussion, we compute and compare effective stiffness of five large-scale digital rocks (up to 10243 voxels) using a variety of numerical approaches, including dynamic pulse propagation method, NIST’s finite element method, and a FFT approach that allows for stress and strain loading boundary conditions. We find that strain loading conditions, using the FFT method, generally lead to stiffer results than those using the stress boundary conditions. Interestingly, the results calculated using the dynamic pulse propagation method agree with those calculated using the static strain boundary condition. Also, we find that laboratory-measured rock stiffnesses are consistently softer than those computed numerically using the dynamic pulse propagation method, the NIST solver, and strain boundary condition of the FFT method. There are many possible reasons for this, including inability of the micro-CT technique to image grain-contacts, uncertainty in stiffness of grain properties, and lack of image clarity to segment soft clay minerals.
Presentation Date: Wednesday, September 27, 2017
Start Time: 1:50 PM
Location: Exhibit Hall C, E-P Station 4
Presentation Type: EPOSTER
ABSTRACT: We compare different algorithms and boundary conditions for Digital Rock Physics modeling. We report results of Simpleware, Comsol and NIST for elasticity and electrical conductivity and Simpleware and Lattice-Boltzmann for permeability. Overall there is good consistency, though there are sample-to-sample differences. The comparison shows a good agreement between Simpleware and Comsol with a difference of ˜1% for effective elastic properties, ˜1-3% for permeability and 5% for effective conductivity. The Simpleware Physics Modules allow a streamlined workflow to compute effective properties and is especially suitable for processing a large number of digital samples. Comsol Multiphysics on the other hand allows more multiphysics coupling and access to internal field variables useful for research purposes. The NIST codes for elasticity and electrical conductivity are very efficient but are specifically only for those two physical properties.
Computation of physical properties on 3-D CT-scanned digital rocks is used in geosciences applications, and a number of different workflows and algorithms have become available. We discuss a comparative study on elasticity and electrical conductivity in different rock types (Figure 1) using Simpleware, NIST (National Institute of Standards and Technology) and Comsol Multiphysics codes. Permeability is also investigated using Simpleware and Lattice-Boltzmann permeability simulators. The 3D CT-scan images of different rock types are Sandstone (S2), Berea Sandstone, Fontainebleau Sandstone, Carbonate (C1) and Carbonate (C2), as well as digitized version of Finney pack with identical spheres (Figure 1). The materials for the grains and pore-fill assumed for the study were water, quartz, and calcite (for the carbonate rocks), with the following elastic properties and conductivities:
2. VOLUME MESHING
2.1. Simpleware (SW)
We import the 3D scan volumes of each sample in Simpleware ScanIP (Synopsys, Mountain View, USA) for image processing and segmentation. Since all of our 3D scan volumes are 100×100×100, each stack has 100 images of 100×100 size. We perform meshing using the Simpleware +FE-Free meshing algorithm with tetrahedral elements. For permeability, the Simpleware Physics Modules +SOLID, +FLOW and +LAPLACE are used with different mesh coarseness Since the rock model is based on a pre-segmented image (a binary image without any greyscale information reflecting the sub-voxel geometry of the sample), it is best to select the “Binarise before smoothing” option. In this way the meshing algorithm in Simpleware does not try to use any spurious greyscale information that might have been introduced into masks by thresholding, and the generated mesh has porosity closer to that of the original image.
Ning, Yang (University of Houston) | He, Shuai (University of Houston) | Liu, Honglin (PetroChina Research Institute of Petroleum Exploration & Development, Langfang Branch) | Wang, Hongyan (PetroChina Research Institute of Petroleum Exploration & Development, Langfang Branch) | Qin, Guan (University of Houston)
The apparent permeability function in kerogen is developed by combining effects of viscous flow, Knudsen diffusion, and surface diffusion, in which the surface diffusion is defined as the transport of adsorbed gas due to the difference of adsorbed gas concentrations. The concentration of the adsorbed gas is quantified by the monolayer adsorption coverage and the specific surface area in shale. Weighting coefficients of viscous flow and Knudsen diffusion are determined based on the probabilities of collisions frequency between gas molecules and between gas molecules and pore walls. Analyzing the effect of surface diffusion, we found that the mass flux contribution of surface diffusion was dependent on many parameters that include pressure, temperature, pore size, specific surface area, etc. The apparent permeability function is further incorporated into the generalized lattice Boltzmann method (LBM) for porous media. A 3D nanometer-scale kerogen digital rock including macro-pores and larger-scale digital rocks including mineralogical components are reconstructed based on the focused ion beam-scanning electron microscopy (FIB-SEM) and Nano-CT experiments, respectively. The kerogen structures (1.53μm3) from FIB-SEM experiments distinguish macro-pores (>10nm) and kerogen solids. Kerogen solid is permeable as micro/meso-pores (1-10nm) exist in nature but are lost by the FIB-SEM experiment. The apparent permeability function is assigned locally on permeable kerogen solids. In Nano-CT digital rocks (413μm3), 4 different mineralogical components have been differentiated. Rock/fluid properties of shales in nanometer scale and micrometer scale are obtained by performing LBM simulations on digital rocks. The relations of methane permeability in kerogen with pressure, temperature, average pore diameter, and specific surface area have been investigated for the FIB-SEM digital rock. In addition, we have investigated the effects of organic contents on methane permeability in shale by comparing the simulation results of 3 different Nano-CT digital rocks.
Ning, Yang (University of Houston) | He, Shuai (University of Houston) | Liu, Honglin (Research Institute of Petroleum Exploration & Development – Langfang Branch) | Wang, Hongyan (Research Institute of Petroleum Exploration & Development – Langfang Branch) | Qin, Guan (University of Houston)
It is well known that shale formations exhibit multi-scale geological features such as nanopores in the formation matrix and natural fractures at multiple length scales. The key challenge in unconventional reservoir simulations is thus how to preserve fine-scale information in coarse-scale reservoir simulations for correct production forecasting and reserve estimation. Accurate prediction of shale permeability using numerical tools requires understanding of transport mechanisms in nano-scale, and upscaling from nano-scale to larger scale simulations. In our recent work (URTeC: #2459219), we presented the coupling of the molecular dynamics (MD) simulation with the lattice Boltzmann method (LBM) on multiple-scale digital rocks to estimate the transport property of shale matrix in micrometer scale. As an extension, this work is aimed to develop an upscaling workflow that integrates nanometer-scale simulations, micrometer-scale simulations and centimeter-scale simulations. The proposed approach allows calculating macro-scale transport properties of natural gas in shales with significantly reducing the loss of critical fine-scale (nano-scale) information.
The reconstructions of multi-scale shale digital rocks are performed using multiple imaging techniques, i.e. FIB-SEM, Nano-CT and Micro-CT. These experiments provide micro-scale pore architectures (~nm), meso-scale mineralogical distribution (~µm), and macro-scale natural-fracture network (~cm), respectively. These multi-scale digital rock reconstructions are then utilized for the investigations of multi-scale transport properties of gas shales. This upscaling process can be summarized as the following three steps. (1) nano-scale transport properties in organic and inorganic structures are calculated using the non-equilibrium MD simulations. Representative organic (kerogen) and inorganic clay (montmorillonite) molecules are built upon their molecular formulas. Transport properties determined from MD simulations are then served as input parameters for LBM simulations in larger scale; (2) micro-scale properties of each component are mapped stochastically on its corresponding voxels in Nano-CT digital rocks. The meso-scale permeabilities of Nano-CT digital rocks are then calculated using the generalized LBM model in porous media; (3) the effective permeabilities of the macro-scale shale digital rock (Micro-CT) with micro-fracture networks are calculated using the generalized LBM model, in which the matrix permeabilities obtained from the step 2 and the transport properties of micro-fractures are served as simulation inputs in macro-scale.